Experiencing the Peer Feedback Activities with Teacher’s Intervention through Face-to-Face and Asynchronous Online Interaction: The Impact on Students’ Writing Development and Perceptions

The objectives of this study were to compare the impact of peer feedback implementation with teacher involvement through training in the classroom and asynchronous online communication on the quality of students’ writing revisions, as well as to investigate students’ perceptions of peer feedback activities. Twenty-five students participated in the experimental study. Eleven students were willingly to be interviewed. Inferential statistical analysis was used to interpret the quantitative data collected from students’ essay writing scores. Meanwhile, the data obtained through observations and interviews was interpreted using qualitative coding analysis. The results of the inferential statistical analysis revealed that peer feedback activities conducted through asynchronous online interactions had more significant effects compared to those conducted face to face on students’ writing revision. Further, after conducting a thematic analysis, six themes emerged: 1) peer feedback activities could increase students’ autonomy in learning, 2)the teacher’s involvement in peer feedback activities was beneficial in terms of improving the consistency of feedback and revision, 3) peer feedback through asynchronous online interactions gave extra time to produce more beneficial comments, 4) peer feedback activities through asynchronous online interactions gave more chances to become a writing audience, 5) communicating via Facebook made the students feel awkward, and 6) recorded feedback via Facebook comments was more beneficial for students’ revision. The implication of the research is that teachers of English needs to consider asynchronous online interactions for students’ writing revision when teaching writing.


Introduction
The process writing approach has been implemented in writing classes since the 1980s. Harmer (2014) mentioned that brainstorming, writing a first draft, revising, editing, and reviewing are the five stages of the process approach to writing. Those steps are commonly conducted in sequence but certain steps can be done repeatedly when it necessary (Hayes, 2012). For instance, the brainstorming stage can be done in two or three repeated stages. Thus, the main goal of the process writing approach actually is not about accomplishing the written text as the product of writing itself but it is more about helping the learners go through each step in the process.

EXPERIENCING THE PEER FEEDBACK ACTIVITIES WITH TEACHER'S INTERVENTION
The sources of feedback could be from the teacher or from peers. Raihany (2014) added that feedback from either the instructor or one's classmates could come in the form of both written and oral feedback. Feedback from students is called peer feedback or what others call peer review, peer editing, or peer revision (Lei, 2017). Both teacher feedback and peer feedback are commonly use when teaching writing in EFL and ESL contexts.
Some researchers have claimed that teacher feedback has not been successful since the students only copy what the teachers revises in their writing without comprehending the intention of the revisions (Pan, 2010;Skipper & Douglas, 2015). The tendency of students to just follow the teacher's instruction is merely because most students believe that the teacher has absolute power where they feel that they have no option but to accept the teacher's authority (Ferris, 2014;Kangni, 2015).
Due to the lack of teacher reviews, peer feedback has become commonplace in writing practices. There are many research studies that consider peer feedback beneficial for students. First, peer feedback in writing activities could foster students' writing academic performance Noorozi & Hatami, 2019). Secondly, Yalch, Vitale, and Ford (2019) asserted that by sharing feedback with their peers, the students become more critical, and by receiving feedback the quality of their writing increased. As a result of sharing comments, students become accustomed to analyzing their peers' writing and recognizing their peers' errors. The activity also made the students aware of not committing on writing mistakes (Patchan & Schunn, 2015). Additionally, by experiencing peer feedback activities, the students' learning autonomy and social awareness developed (Astrid, Rukmini, Sofwan, & Fitriati, 2017;Kuyyogsuy, 2019).
However, some researchers have indicated that not all peer feedback implementation is successful. Peer feedback fails due to several factors. The first is students' inability to give beneficial feedback for their peers' submissions (Bijami, Kashef, & Sharafinejad, 2013;Kangni, 2015). The second is students' cultural factors such as their hesitance to comment on their peers' writing drafts and the tendency to fully trust their teacher's feedback but not feedback from their friends' (Fin, 2018) Finally, time restrictions can limit the number ofcomments other students can make when revising their writing (Al-Bashir, Kabir, & Rahman, 2016).
In order to solve these problems, some experts have proposed teachers' intervention training during peer feedback sessions. The training can be done prior to the feedback session (Hojeij & Baroudi, 2018;Hyland, 2003;Naranjo, 2019;Rollinson, 2005) and during the peer feedback session (Ferris, 2014;Hyland, 2003;Rollinson, 2005). Research has shown that peer feedback training may help students improve the quantity and consistency of the feedback on their peers' writing (Hojeij & Baraoudi, 2018;Min, 2005) and improve the quality of their drafts (Esmaeeli, Abasi, & Soori, 2014;Min, 2006).
Nowadays, the trend to use technology in language teaching is increasing. Th technology used in teaching writing can also be beneficial for developing students' writing performance (Miftah, 2018). One product for technology is the internet. It has been used as a gateway for communicating and learning through the computer. Due to the increased familiarity with the use of computers as a medium for communication, the term CMC (Computer Mediated Communication) has become popular as the internet is used to help students interact during language learning activities. The process of interaction through CMC can be done in two different modes: asynchronous and synchronous communication. Asynchronous CMC is thought to be a good way to improve students' writing skills since the students have more time to compose, provide feedback on their peers' essays, and reflect on theirdrafts (Balaji & Chakrabarti, 2010;Kitade, 2008;McCarthy, 2010). The importance of online learning processes, especially in teaching English at schools and universities, has been growing since the COVID-19 pandemic began, especially so in Indonesia.
Facebook is a popular social media platform in Indonesia. By 2020, Indonesia's Facebook user base grew to 140 million (Jemadu & Prastya, 2021). The use of Facebook has also become familiar with English teachers and many of them have used Facebook as a learning medium when teaching English writing. Several researchers have published on the advantages of using Facebook for writing practice where by having Facebook writing activities, students' writing performance was improved (Ping & Maniam, 2015;Yunus, Salehi, & Chenzi, 2012).
Only a few research studies have compared how students' writing draft quality improved from their first draft to the final revision as a result of exchanging feedback (Astrid & Marzulina, 2018;Hidayat, 2020;Moloudi, 2011). However, no study has compared the consistency of students' revisions in terms of writing elements such as substance, coherence, grammar, word selection, and mechanics, between classroom and asynchronous online communication peer feedback, and there is no research study which compares students' perceptions after experiencing those two kinds of peer feedback interactions.
As a result, we wanted to conduct a study to compare the impact of implementing peer feedback activities with teacher's training through face-to-face and asynchronous Facebook interactions on the quality of students' drafts and to explore their perceptions of those peer feedback activities.

Literature Review
As a crucial element of the process writing approach, feedback provides input given by a reader to the writer to help them revise any mistakes or point out any confusing sections found in the writing draft . As a result of sharing comments and doing revisions, the students' understanding of writing mistakes that are likely to be made on the writing draft can increase. Feedback can be in the form of criticism, suggestions, or even questions to consider when revising the draft.
There are two sources of feedback: feedback from teachers and feedback from students. The two forms of teacher feedback are written feedback and oral conferences. There are three types of teacher's written feedback: requests, criticism, and praise (Zahidah, Farrah, & Zaru, 2014). Each type of feedback has a different focus. The feedback can focus on language aspects, content, genre, and the process of writing (Hyland, 2003). Further, oral feedback is given in the form of an oral writing conference. It is a form of consultation, where the teacher can give feedback orally and the students could consult on their writing with their teacher. In this session, the students have an opportunity to clarify any written feedback given on their essay that they did not understand.
Peer feedback is a type of feedback that is provided by peers in which students are responsible for providing corrections to their peers' essays in groups. The action takes place in a small group consisting of four to five students. In the group, the students are required to read each of their peers' drafts and share comments. After sharing feedback, the studentsrevise their writing by employing those comments (Cao, Yu, & Huang, 2019;. Since the implementation of peer feedback is not always successful, some researchers have proposed the teacher's intervention, which involves the teacher giving the students training prior to the peer feedback sessions (Naranjo, 2019;Rollinson, 2005). The main goals of pre-training session are first to increase students' recognition on the intention of the peer feedback activities, second to make students ready to convey their thoughts in a group properly and to practice the appropriate language used in feedback sharing sessions, and finally to illustrate the way to give beneficial feedback and the way to manage the feedback when revising.
Intervention training should be given to the students in the middle of the writing process, (Rollinson, 2005). The goals of the training are to increase the effectiveness of giving feedback to peers' writing. In this session, the process of peer feedback that has been conducted is reviewed. Then, mistakes frequently found during the activity related to the way of giving feedback and revising the writing drafts are discussed. Following that, several recommendations are made in order to increase the efficiency of the reviewing sessions.
Rather than having a teacher intervene in peer feedback via direct contact in the classroom, peer feedback can be done online through computer mediated communication (CMC). CMC peer feedback is a promising way to improve students' writing skills (Latifi, Noorozi, Hatami, & Biemans, 2019). The use of online media for conducting peer feedback is becoming popular when teaching writing since it creates an environment where the students can write, share feedback, and make revisions without time and space restrictions (Latifi et al., 2019;Noorozi, Biemans, & Mulder, 2016;Noroozi et al., 2020). Guidance on giving feedback could also be embedded in such an online environment to help students provide more relevant feedback (Latifi, Noorozi, & Talaee, 2021) When implementing online peer feedback, the learning activities are done through asynchronous interactions where the communication does not occur at a certain time and place (Faja, 2013;Veldhuis-Diermanse, 2002, as cited in Espasa, Guasch, & Alvarez, 2013). Social media sites such as Facebook can be used as a communication platform (Boas, 2011). Facebook is very comfortable to utilize since it provides space to write as many characters as we want, to share feedback as a comment, and to send pictures and videos. Facebook also allows users to form groups .

Research Design
An embedded mixed methods design was used in this research (Creswell, 2012).During the experimental study, all kinds of data were obtained simultaneously. The embedded mixed methods design used in this study is depicted in the diagram below:

Figure 1
Visual diagram of embedded mix method design procedure

INTERPRETATION
The design used for the experimental phase was a repeated measures design. In this research, there was only a group of students. The students were handled by using peer feedback in writing activities with two types of interactions: synchronous in-class interactions and asynchronous online interactions through Facebook. The students' writing performance after the first treatment (exchanging feedback with face-to-face interactions) was compared to their writing performance after the second treatment (peer feedback with online interactions).
The repeated measures design is illustrated below.

Figure 2
Visual diagram of the repeated measures design

Research Setting and Participants
The research was conducted in a writing class at an Islamic university in Palembang, South Sumatera, Indonesia. Twenty-five students participated in this research. The participants were fourth-semester students. Their average age is 21. The students were from the class of one of the writers, who was their writing lecturer and knows well their prior writing ability. The head of the study program allowed the class to be observed and treated during the research. Those 25 students joined an essay writing class as one of the prerequisite subjects to be taken in the fourth semester. For the sake of their rights and research ethics, the students were told in advance that their identities would be kept anonymous. Then, the students were asked to fill out the consent form on their willingness to become the participants of the research. All 25 students submitted their consent form to become participants of the experimental phase of the research.Out of the 25 students, 11 students were willing to be interviewed.

Data Collection
After the treatments, data in the form of students' revision writing scores was collected in the quantitative process of data collection. The experiment was done in 16 writing class meetings by implementing face-to-face peer feedback with teacher's intervention (Naranjo, 2019;Min, 2006;Yang, Badger & Yu, 2006;Rollinson, 2005) and asynchronous online interactions (Annamalai, 2018). Firstly, the students were taught by implementing face-to-face peer feedback. They were taught how to brainstorm and outline their ideas. After that, they were required to write their first writing draft. Then, the students were grouped into groups of four in order to have peer feedback sessions. Prior to the sessions, the students were trained on how to provide appropriate feedback in terms of communicating their feedback to their peers, responding to peer feedback, and providing useful feedback in terms of writing aspects (writing substance, paragraph organization, diction, grammar, and mechanics). The students had their first peer feedback session after the pre-training session. Based on their peers' comments, the students were asked to rewrite their first essay's draft. Following that, the lecturer provided intervention training. The training was aimed to help students develop their ability to provide useful writing feedback. After the training, the students were invited to engage in a second peer feedback session. In this second session, the students had to revise their draft as a final revised draft. After all the steps of the faceto-face interaction were completed, the next peer feedback activities were conducted through asynchronous online interactions by employing Facebook as the medium of communication.
The final revision drafts from both the direct classroom and the Facebook interactions were scored to compare the students' writing performance after receiving comments. To reduce subjectivity, two independent raters from another university were asked to score the final drafts. The writing rubric proposed by Hyland (2003) was used as the guideline to score the students' writing. The rubric consists of three scoring criteria of writing aspects: (1) format and content (FC); (2) organization and coherence (OCh) and (3) sentence construction & vocabulary (SCV).
The qualitative data were collected as supporting data during the qualitative process. This was done to explore how the students perceived the peer feedback activities. In this qualitative phase of data collection, interviews and observations were employed. Observations were done during meetings to see if the students participated in peer feedback events, either face to face or through Facebook interactions. All of the phenomena that took place during the observations were noted. After all of the treatments, the students were interviewed to find out how they felt about the learning experiences.

Data Analysis
Parametric tests were used to evaluate the quantitative data obtained from students' final draft scores. Parametric tests were used since it was assumed that the shape or variance of the writing scores in the population were normally distributed. Multivariate Analysis of Variance (MANOVA) was the inferential statistic. SPSS 22 was used to perform the statistical analysis (Santoso, 2014).
Meanwhile, the qualitative data from the observations and the results of the students' interviews were interpreted using coding analysis (Heigham & Croker, 2009;Heriyanto, 2018). The steps of coding analysis were as follows: first, before any coding can begin, the data were organized by data grouping; then, the collected data were reduced, where the data to be analyzed were selected based on the research questions; next, the data were read thoroughly; then the codes were assigned to represent the students' perceptions; finally, the data were coded and refined to detect emerging themes.
Moreover, the data were checked in terms of their trustworthiness as a requirement of qualitative studies (Creswell, 2012;Heigham & Croker, 2009). In order to assure the credibility of this study, prolonged and persistent observations were conducted. After interviewing the research participants, a member checking method was used. In this session, the respondents were required to re-read the interview transcript to ensure that the contents of the interview transcript were appropriate to what the respondents had shared during the interviews. Finally, data triangulation was employed, where the qualitative data were obtained from multiple sources of data, the observation field notes and interviews.

The Impact of the Face-to-Face Peer Feedback Compared to that of Asynchronous Online Interactions
The results of incorporating face-to-face and asynchronous online contact peer reviews were compared using the MANOVA statistical tool. The aim of the study was to compare the impact of implementing face-to-face and asynchronous online interactive peer feedback activities on the quality of students' revision drafts in terms of writing aspects such as format & content (FC), organization & coherence (OCh), and structure & vocabulary (SV). The results of the analysis are displayed in Table 1. According to the results of the multivariate analyses, all of the sig. values from each test were 0.00, which is less than 0.05. As a result, the impact of direct and asynchronous Facebook peer feedback on the students' writing performance differs significantly.
Finally, tests of between-subjects effects were used to see whether there was a significant difference in the influence of each implementation of peer feedback interactions on each element of writing. Table 2 shows the outcome of the analysis. The data listed in Table 2 points out that the sig. values for the aspects of Format & Content (FC) and Sentence structure & Vocabulary (SV) were 0.00, which are lower than 0.05, meanwhile the sig. value for the aspect of Organization & coherence (OCh) was 0.893, which is higher than 0.05. Thus, there was a substantial difference in the impact of implementing peer input by face-to-face and asynchronous online interactions on the aspects of format and content, sentence structure, and vocabulary, but not on the aspect of organization and coherence. The results of each peer feedback implementation on each writing element are compared in Table 3.
The data listed in Table 3 revealed that the implementation of asynchronous Facebook interaction peer feedback had more of an effect more on the aspect of FC and SV compared to that of face-to-face interactions. The influence of these two peer feedback practices on the aspect of OCh, however, did not vary significantly.

Students' Perceptions toward Face-to-Face and Asynchronous Online Interactive Peer Feedback Activities
The results of the coding analysis revealed how the students viewed peer feedback practices in both face-toface and online interactions. Seven themes that emerged and they are described below.

Both Peer Feedback Activities Could Increase Students' Autonomy In Learning
It was apparent from the observations and interviews that students had become self-sufficient in their learning. This was because the peer feedback sessions encouraged the students to become more independent by trying to identify their peers' mistakes in writing and give feedback for revisions.
Based on the observations done in the class during peer feedback activities, it can be stated that all of the students were actively engaged in the process of sharing comments through both direct and online interactions. All of the students in the groups for peer discussions could share their feedback after reading their peers' writing. Most of students could identify mistakes in their friends' drafts and make corrections related to those mistakes. This also happened when the same students were required to give peer feedback through Facebook. The students actively gave feedback to their peers' drafts in groups and they did it on time based on the schedule.
Moreover, it was clear that the students became independent when they were required to undertake the peer feedback activities. They read their friends' essays seriously, tried to identify mistakes, and tried to give beneficial comments for corrections.One of the students named M clarified in the interview, "When I was required to give feedback on my friends' writing, firstly I was awkward, since I know that my capability in writing is bad, but I tried hard to give beneficial comments." Another student named A added, "When I was asked to read and to give feedback to my peers' writing, I read those essays carefully. I identified the mistakes and corrected those mistakes. In order to make sure that my corrections were appropriate, I re-read my books and consulted my dictionary."

The Lecturer's Involvement in Peer Feedback Activities Was Beneficial in Terms of Improving the Consistency of Feedback and Revision
We learned from the interviews and observations that the lecturer's intervention, through both face-to-face and asynchronous online interactions, was well executed. The students thought that the lecturer's presence during peer feedback activities, both through direct and asynchronous online contact, was helpful in guiding them to exchange feedback with their friends and update their essays based on input from their colleagues in the group. For example, a student named Y said, "The way my lecturer guided us on how to conduct peer feedback was clear. I got the idea on how to give feedback to my friends' essays and revised mine as well." Another student named J added, "From intervention training I could realize what I had done wrong related to the content of my comments to my friends and the way to revise my essay. After the instruction from my lecturer, I did not do the mistakes again."

Peer Feedback Through Asynchronous Online Interactions Gave Extra Time to Produce More Beneficial Comments
After analyzing the data, it became apparent that asynchronous interactions through Facebook allowed the students to have more time to read and give comments to their peers' writing. The extra time gave the students chances to read their peers' writing in detail and offer complete corrections. As a result, the students' feedback on Facebook appeared to be more informative than the feedback provided through face-to-face communication.
Furthermore, we learned that most students felt free to give comments anytime and anywhere through asynchronous online interactions. They were not forced to do peer feedback activities in a limited amount of time. For example, one student named T said in the interview session, "When I was asked to read and give comments to my peers' drafts through Facebook, I could determine when and where I could do those activities. I had more space to read the essays. I even could manage to read the essays more than once. By reading carefully, I felt I could give more comments that I feel were more beneficial compared to those I gave when I was required to do peer feedback activities in the class." Another student named E clarified, "I felt that I could not be maximal in giving feedback to my peers' essays because the time was very inadequate. I did not have enough time to read my peers' essay thoroughly. That was why sometimes I just gave general comments or even I could not recognize the mistakes made by friends in their essays." In addition, a student named R added, "When the process of giving feedback was conducted through Facebook, I felt I could perform well. After identifying my friends' mistakes, I had enough time to consult my books as references to make sure whether my feedback was already okay or not." Additionally, it appeared that some students felt uncomfortable with the noisy class situation. Some of them preferred a quiet situation to perform better. As a result, the comments given were not maximal. Most of comments were related to grammar and mechanics. There were rarely comments about the content of the essay and about coherence. For example, one student named B mentioned in the interview, "I lost my concentration whenever I was in such a noisy situation. Thus, I could not get the message the writer intended to say. Since I was not in the mood to read, I only gave comments related to the mistakes my friends made in terms of grammar, spelling, and punctuation." The information we got from the interviews was in line with that we learned from the observations. From the observations, we found that everything ran well while peer feedback was being conducted in the class. However, the activities seemed chaotic. The class became noisy. Peer feedback activities in the class seemed to be conducted in a rush. Students were instructed to give comments for at least three of their friends' essays within a limited time, about two hours. This limited time period made the process of reading and giving comments unfavorable.

Peer Feedback Activities Conducted through Asynchronous Online Interactions Gave More Chances to Become a Writing Audience
The results of the data analysis revealed that peer feedback sessions, whether performed face to face or asynchronously through Facebook, provided students with the opportunity to read their friends' essays in a group setting. However, some students acknowledged that they had more time because they were receiving reviews from their peers via Facebook. They had the opportunity to read many more essays from their friends, even those who were not in the same group. Students can learn how to write, recognize errors, and learn not make the same mistakes as their friends by reading more essays shared on online.
In the interview session, one student named N clarified, "In the Facebook sessions, I was required to read and share feedback on my peers' essays in the group. However, I was curious to read my other friends' essays since they were posted openly. From reading those essays, I could get the benefit such as learning the patterns of a good essay from the essays which I considered better." Another student named M said, "I also found that there were some mistakes made by my friends in their essay. By identifying the mistakes I was aware not to make the same mistakes."

Communicating Via Asynchronous Online Interactions Made the Students Feel Awkward
We can see from the results of the data analysis that the students felt awkward when they were required to communicate with their friends and lecturer through Facebook during peer feedback activities. From the interviews, most of students admitted that they preferred direct communication to online communication. One student named A said, "I felt reluctant to give feedback to my friends because I could not see their faces directly. By seeing their faces I can be sure whether they could understand what I said to them or not." Another student named C said, "I preferred to have direct communication with my friends and my lecturer. When I gave feedback to my peers in the discussion session, I could catch their facial expressions that could give me clues whether or not they understood what I said to them." Finally, one student named R said, "I prefer to have my lecturer's explanation through face-to-face interactions compared to online interactions. This was because I could see her face directly when I asked a question to her. This made me much more comfortable."

Recorded Feedback Via Facebook Comments Was More Beneficial for Students' Revisions
Based on the information we got from the interviews, most students took advantage of the feedback given by their friends and recorded in comments section. Since all of the comments were recorded, it was beneficial for the students to re-read the comments any time they wanted to revise their essays. One student named S said, "The feedback given from my friends recorded in the Facebook section was beneficial. I could crosscheck the comments whenever I wanted to revise my essay." Then, one student named A said, "Sometimes I could not catch the feedback from my friends while we had oral sessions in peer feedback activities. Then, when I had to revise my essay, I could not even read my notes since I wrote them in a hurry, thus my revision was not maximal."

Discussion
In terms of format and content, as well as sentence structure and vocabulary, the consistency of the students' revisions was evident after completing peer feedback activities through asynchronous online interactions compared to those conducted face to face, according to the study. This was caused by the opportunity given to the students to have more space and time to give feedback and to do revisions. This is in line with Balaji and Chakrabarti (2010), Kitade (2008), and McCarthy (2010), who mentioned that by conducting peer feedback activities through asynchronous online interactions, students could decide when they had time to give feedback and responses to their friends. This could not happen when they were required to carry out peer feedback activities in the class, where it seemed like they were forced to do everything simultaneously.
It was also believed that the teacher's participation in peer feedback practice, whether in face-to-face or asynchronous online interactions, was beneficial to the students. They acknowledged that from the training given by the lecturer, the students understood the way to give beneficial feedback to their peers' writing draft, the guidelines for writing aspects that must be commented on, and the way to communicate well in group discussions. The review session in the middle of the activity was also considered beneficial to help make sure everything ran well. This fact relates to Hyland (2003), who asserted that the intervention from teachers in peer feedback activities is considered important since the students in the EFL context commonly have lower language competence.
Moreover, the research illustrated that by having a longer time and wider space to give comments, the students had more chances to read their friends' essays thoroughly and considered the feedback more detailed and clearer compared to when they were asked to share comments among their friends in a limited amount of time. This is in agreement with the research carried out by Aydawati, Rukmini, Bharati, and Fitriati (2018) stating that peer feedback activities organized in online environments made the students more focused. As a result, despite only giving corrections on local aspects such as grammar and mechanics, students provided more comments on global writing elements like content and ideas as well as paragraph organization (Jones, Garralda, Li, & Lock, 2006;Liu & Sadler, 2003).
The students also perceived that peer feedback activities could give them more chances to become a writing audience. Becoming a writing audience as one advantage of peer feedback has been asserted by Rollinson (2005). He said that by being aware of becoming a writing audience when the students are required to do peer activities, they will become critical readers. As the result, they will produce more critical comments and be able to revise their friends' essays, and even their own. Since Facebook is a social media platform where everything posted could be seen by everyone in the group, the students had more opportunity to become readers of their friends' essays compared to during in-class activities.
According to the findings of this study, peer feedback, whether given face to face or through asynchronous online interactions, may improve students' learning autonomy. To put it another way, the students were independent in their learning. Apriani et al. (2020) claimed that peer feedback activities were advantageous thanks to the freedom to make decisions during the learning process. As a result, students' writing abilities will improve. Shih (2011) claimed that learning in an online environment lets the students become independent to learn and acquire knowledge from many sources provided online.
Further, the students in this study admitted that they could read the comments from their friends any time they wanted to revise their essays because they were recorded in the comments section on Facebook. This is in line with Balaji and Chakrabarti (2010), who mentioned that in online interactions, comments are stored permanently and that makes them easier for students to access.
Some researchers have claimed that peer feedback activities failed due to the cultural backgrounds of their students, especially in Asia. Carson and Nelson (1994) as cited in Levine, Oded, Connor, and Assons (2002) discovered that students' cultural backgrounds influence the way they give feedback. This was found to be the case for Chinese studentsgiving peer feedback in an ESL class (Yang, Badger, & Yu, 2006). Peer feedback was unsuccessfully implemented because the students were unable to criticize their peers' writing. However, in this research, most of the students claimed that they were more comfortable communicating with their friends and lecturer face to face compared to through Facebook. This was because by having communication face to face they could see their peers' facial expressions directly when they had to give feedback and it made them much more comfortable. Thus, the important benefit of blended learning the ability to combine face-to-face and online interactions. As Kintu, Zhu, and Kagambe (2017) mentioned, blended learning is a concept that integrates the learning process through synchronous interactions in the class and online asynchronous interactions.
The comments section provided in Facebook saved all of the students' comments. This made it easy for the students to re-read all the comments when revising their essays. This is in line with Balaji and Chakrabarti (2010), who said that there is no data loss for online direct feedback since published posts from students are saved virtually where they can be accessed and read at any time.
To conclude, peer feedback has the ability to improve students' writing skills if the tasks are well-managed in terms of how to group the students and how to organize the feedback sharing session and revision activities. Thus, even though peer feedback activities are centralized on students, the roles of the teachers cannot be neglected. In the middle of a sharing feedback session, teachers still have to monitor the activities in order to ensure that everything runs properly. Finally, writing in an online environment was shown to alleviate students' tension in writing. However, face-to-face communication either with the teacher or with their peers cannot be neglected.

Conclusion
There were two conclusions of this study. First, peer feedback activities conducted through asynchronous online interactions had a more significant effect compared to those conducted face to face on students' writing revision. The second, students gave more positive responses to peer feedback activities conducted through asynchronous online interactions than face to face. There were six themes based on the thematic analysis to address the second objective. Those were 1) peer feedback activities could increase students' autonomy in learning, 2)the lecturer's involvement in peer feedback activities was beneficial in terms of improving the consistency of feedback and revision, 3) peer feedback through asynchronous online interactions gave extra time to produce more beneficial comments, 4) peer feedback activities through asynchronous online interactions gave more chances to become a writing audience, 5) communicating via Facebook made the students feel awkward, and 6) recorded feedback via Facebook comments made it more beneficial for the students' revision. Finally, we realize that the number of participants in this study was small. This was because we would prefer to investigate the issues more deeply from a smaller number of the students. As a result, it is suggested that future researchers conduct similar studies with a larger sample size and a broader scope.